| from config import TTS_MODEL, CHAR_LIMIT, configure_tts |
| import re |
| import wave |
| import tempfile |
| import os |
|
|
| tts_client = configure_tts() |
|
|
| def temp_wave(): |
| fd, path = tempfile.mkstemp(suffix=".wav") |
| os.close(fd) |
| return path |
|
|
| def chunk(text, limit = CHAR_LIMIT): |
| words = text.split() |
| chunks, cur = [], "" |
| for w in words: |
| if len(cur) + len(w) + 1 <= limit: |
| cur = (cur + " " + w).strip() |
| else: |
| if cur: |
| chunks.append(cur) |
| cur = w |
| if cur: |
| chunks.append(cur) |
| return chunks |
|
|
| def stitch(paths): |
| if len(paths) == 1: |
| return paths[0] |
| out = temp_wave() |
| with wave.open(paths[0], "rb") as w0: |
| nchannels = w0.getnchannels() |
| sampwidth = w0.getsampwidth() |
| framerate = w0.getframerate() |
| with wave.open(out, "wb") as o: |
| o.setnchannels(nchannels) |
| o.setsampwidth(sampwidth) |
| o.setframerate(framerate) |
| for p in paths: |
| with wave.open(p, "rb") as w: |
| o.writeframes(w.readframes(w.getnframes())) |
| return out |
| |
| def synthesize(text, voice="troy"): |
| try: |
| paths = [] |
| for c in chunk(text): |
| p = temp_wave() |
| resp = tts_client.audio.speech.create( |
| model=TTS_MODEL, voice=voice, input=c, response_format="wav", |
| ) |
| resp.write_to_file(p) |
| paths.append(p) |
| return stitch(paths) if paths else None |
| except Exception as e: |
| print("TTS failed:", e) |
| return None |
